Asked about Retries, rate limits and timeouts

You are calling an API 1000 times and keep hitting rate limits. What do you do?

An answer that works

Cap concurrency with a semaphore rather than firing everything at once, read the rate limit headers and slow down before being rejected, back off exponentially with jitter and honour retry_after, and use batch endpoints for anything that can wait.

Reading an answer is not being able to give one. The section that teaches this is Retries, rate limits and timeouts, 5 hours of HTTP & APIs.

Learn this properly

Also asked about HTTP & APIs

What is the difference between 401 and 403?Which HTTP errors should you retry?What is idempotency and why does it matter for API calls?Why add jitter to exponential backoff?What does a timeout tell you about whether the server acted?